Identifying Possibly Pleiotropic Comorbidities Using Genetic Variant and Phenotypic Data

For patients suffering from disease, comorbidities and multi-morbidities often complicate treatment and decrease quality of life. Comorbidities, e.g. obesity and diabetes, share known physiological links, while multi-morbidity is defined as the co-occurrence of two distinct disease states. Some multi-morbidities, however, may share hidden physiological or genetic links, and are actually also true comorbidities. It is widely recognized that a single gene can affect multiple phenotypic traits, i.e. physical manifestations. The term “pleiotropy” describes these relationships, where a single gene or gene variant can result in a widely distributed cascade of physical effects. A well-recognized example is the genetic variant linked to sickle cell anemia, which also confers malaria resistance. The identification of these pleiotropic relationships has the potential to impact patient screening and treatment strategies for multi-morbid diseases, as well as provide motivation for targeted research into their shared pathways for new therapies. In this study, we developed and tested a method for identifying possible pleiotropic relationships between diseases. The method consists of the following steps: 1) querying the National Center for Biotechnology Information (NCBI) database ClinVar for all genetic variant records indexed with a specific disease/phenotype. 2) Identifying possible comorbidities from the list of additional disease/phenotypes associated with the original query. 3) Testing for a statistical comorbidity relationship between the original disease and an associated phenotype using nationwide hospital discharge data. To test this method, we queried ClinVar using the disease phenotype “colorectal cancer”. From the list of results, we identified a potential genetic relationship between age-related macular degeneration (ARMD) and colorectal cancer. We then analyzed nationwide hospital discharge data to determine if there is a significant relative risk and odds ratio for patients suffering from both diseases when diagnosed with either ARMD or colorectal cancer. Our data source was the 2012 National Inpatient Sample (NIS) of the Health Care Utilization Project (HCUP) and all statistical analysis was performed using SAS version 9.4. The results of our initial test did not show statistical evidence of a comorbid relationship between the ARMD and colorectal cancer phenotypes. Nevertheless, we believe the method can be applied in the identification of unknown comorbidities. Further, we propose this methodcan prioritize further study into possible pleiotropic relationships; our first test is an example of a relationship that may not be worth prioritizing. Such a method can have direct impact on disease prevention and treatments and increase the speed of future discoveries